Integrated biometrics for human identification integrated biometrics

The new trend in human biometrie authentication is the development of multi-biometric system, preferably using multi-modal signals. A good strategy could be to combine ECG and EEG because of their liveliness and robustness against falsification. In this study, we propose a multi-biometric authentication method for human identification using signals from low-cost devices. EEG signal is first preprocessed using median and bandpass FIR filter to remove noise and artifacts. The baseline wandering effect of the ECG signals is tackled using median subtraction method. Every half of each signal is divided into segments with 90% overlapping. Multiscale wavelet packet decomposition is then applied to each segment and a feature vector, namely wavelet packet statistics (WPS), is computed. Features from ECG and EEG segments are combined using a feature level fusion technique. The combined feature is finally used to train a supervised error-correcting output code multiclass model (ECOC) using support vector machine (SVM) classifier, which ultimately can recognize humans from the disjoint test EEG segments. A preliminary experiment with 10 EEG records from 10 subjects shows 82.9% F-score of the proposed method.